Growth capital investments for vertical software. General partner at Lambda Capital.
Palo Alto, CA
Joined January 2021
- Tweets1.6K
- Following1.1K
- Followers1.6K
- Likes22.6K
I signed up for the X account, the unified X, Cursor, and Grok Bot account.
For the first time ever, I'm completely out of usage in less than a week.
They say I'm on the Super plan too.
It's shocking to me that by unifying the account, my user experience got worse.
@elonmusk please fix this
Meta's Hyperion leases are only 4 years. So Meta is only on the hook for 4 years of rent, right?
Wrong.
The $27.3B of bonds aren't on Meta's balance sheet. But the 10-K shows a $45.95B maximum exposure to loss.
Off balance sheet isn't off the hook.
lambdavp.com/blog/meta-hyper…
SpaceX closed $40B of GPU debt. Most neoclouds can barely borrow right now.
Lenders aren't underwriting the chips. They're underwriting revenue that shows up either way.
Same reason I read signed contracts and renewals, not who led your round.
lambdavp.com/blog/neocloud-f…
Pretty much.
As of the past few weeks, debt markets have almost entirely shut off access to money for any company that doesn’t have substantial revenue independent of AI data centers.
SpaceX closed $40B of debt for GPUs, but we’re investment grade and have Starlink & launch revenue.
That means the 200+ neoclouds are going to see guppy-level death and consolidation. Some will make it through, but not many.
"If France calls upon me."
Meanwhile:
"I saw the crown of France lying in a gutter, so I picked it up with my sword." - Emperor Napoleon I
“The situation has never been so serious; the Fifth Republic is on the verge of collapse. My family has served France for centuries, and if France calls upon me, I will be at its service. The indispensable condition is that France desires the return of the monarchy - a monarchy above parties, unifying.”
- Luís de Bourbon, direct descendant of Saint Louis IX
Domenic Giovanzana retweeted
Folks, the Greek ambassadorship is open. President Trump, I know you love a dark horse candidate. Especially one who has been unjustly persecuted by the Obama DOJ.
If you want someone who knows Athens like the back of his hand from hunting terrorists, I’m your guy.
youtu.be/DRBJ-2HEz-4?is=abeG…
If your retention is strong and your contracts are signed, there are cheaper ways to get the cash today without discounting your own revenue.
I’ve pass on deals that work on paper.
Not because the business model was wrong. Because the founder gave fraud energy.
I once met with a guy where everything in his business looked really good: great revenue, scale, great free cash flow margin, etc. but every time I met with him something felt a little off.
I did some background due diligence on him.
Turns out he was a co-conspirator in a case where he was ripping off old people on mortgage fraud in the amount of $51 million.
Always listen to your gut.
You can never do a good deal with a bad person.
Lying.
The unfortunate nature of fundraising: you can promise everything, deliver nothing, and it doesn’t matter.
Founders need money, and investors don't do due diligence on things such as pipeline size.
This incentivizes founders to lie and overstate their pipeline.
Which leads to more money raised.
And when the founders underdeliver, there's no consequences.
It's not the way things ought to be, but investors should be aware.
Ask: "The $2M deal with ABC co., what stage is that?" and then call that customer to validate what the founder said.
My deal sourcing is autonomous.
That’s how I evaluated 42,000 companies with AI this year.
Scale is the easy part. Trust is not.
While interviewing for this role (below), a candidate said he’d built an AI dashboard for a prior employer that, “helped them make marketing decisions.”
I asked: “How do you know the AI fed the dashboard the correct data?”
He didn't have a good answer.
At Lambda we make real decisions off tools that AI helps write. AI is useful. It is also frequently wrong. If you don’t run realistic tests, the dashboard can look polished and still be lying to you. Then you’re setting strategy on bad numbers.
Simple example: you know 1 + 2 + 3 = 6. Feed the system 1 + 2 + 3. If it returns 15,546, something is broken.
Most people skip that test. Then their vibe-coded slop steers the business way off course.
I cancelled my Claude subscription.
The answers became frustratingly complicated, overly technical, and quite pessimistic.
Grok 4.6 has really impressed me. Among Grok Bot, Cursor, and the core chat feature, I am very bullish on SpaceXAI.
I told a grown man, "Let's date before we get married."
Sounds silly.
I like to loan money to companies before exploring an equity investment.
Often that means a loan first for 6 to 12 months, so both sides can see how it feels before anyone locks into equity. This helps me see if the working relationship holds up and if we're the right partner for the company. If it's a fit, we can talk a permanent partnership. If it doesn’t, nobody’s stuck in a bad marriage.
"Better than a Harvard MBA."
Had an offboarding chat with one of our interns this week.
I told him that a couple years at the right firm can feel like a paid MBA. Not just because of the brand on the resume, but because of the people sitting next to you who go on to do hard, interesting work.
Imagine spending 2-3 years at SpaceX, OpenAI, or Anthropic. That network will stay with you for the rest of your career.
